Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.
TH EHAMILTON PROJECT                               Advancing Opportunity,                               Prosperity and Gro...
The Hamilton Project seeks to advance America’s promise ofopportunity, prosperity, and growth. The Project’s economicstrat...
TH E                HAMILTON             PROJECT                                                              Advancing Op...
Pay- As- Y ou-Drive A uto Insurance	Abstract          The current lump-sum pricing of auto insurance is inefficient and in...
Pay-A s -Yo u - D r i v e A u t o In s u r a n c e	Contents    1. Overview	                                               ...
Pay- As- Y ou-Drive A uto Insurance	        THE H AMILTON PROJECT    |   the broo king s ins titution
Pay-A s -Yo u - D r i v e A u t o In s u r a n c e1. OverviewI    f you are like most Americans, you probably eat        d...
Pay- As- Y ou-Drive A uto Insuranceage-based insurance premiums. Second, the federalgovernment should increase the funding...
Pay-A s -Yo u - D r i v e A u t o In s u r a n c e2. The ChallengeE        ver since Henry Ford made cars affordable to   ...
Pay- As- Y ou-Drive A uto InsuranceFigure 1Yearly Accident Claims by Annual Mileage                           0.10        ...
Pay-A s -Yo u - D r i v e A u t o In s u r a n c e   Figure 2   Accident Claims per Million Miles by Annual Mileage       ...
Pay- As- Y ou-Drive A uto Insurance    Second, lump-sum pricing leads to an inefficiently                                 ...
Pay-A s -Yo u - D r i v e A u t o In s u r a n c esions since World War II. Oil wealth strengthens        cident because t...
Pay- As- Y ou-Drive A uto Insurance3. A New ApproachA         relatively simple and pragmatic way to ad-                  ...
Pay-A s -Yo u - D r i v e A u t o In s u r a n c eA. Designing PAYD Insurance                               various public...
Pay- As- Y ou-Drive A uto InsuranceTable 1Cost of Commercially Available Telematic Mileage-Recording Devices	 Manufacturer...
Pay-A s -Yo u - D r i v e A u t o In s u r a n c eOther models of monitoring mileage exist. For             than fifteen t...
Pay- As- Y ou-Drive A uto Insurancepoints out in the Washington Post (“For Hybrid                       ny. These monitori...
Pay-A s -Yo u - D r i v e A u t o In s u r a n c eties arising from PAYD that cannot be captured by                       ...
Pay- As- Y ou-Drive A uto Insurancelose coverage if they exceeded the prepaid number                        such competiti...
Pay-A s -Yo u - D r i v e A u t o In s u r a n c e    lected period based upon operator driving char-                     ...
Pay- As- Y ou-Drive A uto Insuranceknown in the field of endeavor . . . can provide a                   devices in vehicle...
Pay-A s -Yo u - D r i v e A u t o In s u r a n c e4. A Three-Part Strategy to Encourage PAYD AdoptionI   t is hard to tell...
Pay- As- Y ou-Drive A uto InsuranceCongress should reauthorize the VPP when the                              Given that th...
Pay-A s -Yo u - D r i v e A u t o In s u r a n c epolicies, leading a few more drivers to adopt PAYD.A virtuous circle wil...
Pay- As- Y ou-Drive A uto Insurance5. Impacts, Costs, and BenefitsT       he net social benefits if PAYD were universally ...
07 payd bordoffnoel
07 payd bordoffnoel
07 payd bordoffnoel
07 payd bordoffnoel
07 payd bordoffnoel
07 payd bordoffnoel
07 payd bordoffnoel
07 payd bordoffnoel
07 payd bordoffnoel
07 payd bordoffnoel
07 payd bordoffnoel
07 payd bordoffnoel
07 payd bordoffnoel
07 payd bordoffnoel
07 payd bordoffnoel
07 payd bordoffnoel
07 payd bordoffnoel
07 payd bordoffnoel
07 payd bordoffnoel
07 payd bordoffnoel
07 payd bordoffnoel
07 payd bordoffnoel
07 payd bordoffnoel
07 payd bordoffnoel
07 payd bordoffnoel
07 payd bordoffnoel
07 payd bordoffnoel
07 payd bordoffnoel
07 payd bordoffnoel
07 payd bordoffnoel
07 payd bordoffnoel
07 payd bordoffnoel
Upcoming SlideShare
Loading in …5
×

07 payd bordoffnoel

1,277 views

Published on

  • Be the first to comment

  • Be the first to like this

07 payd bordoffnoel

  1. 1. TH EHAMILTON PROJECT Advancing Opportunity, Prosperity and Growth DISCUSSION PAPER 2008-09 J U LY 2 0 0 8 Jason E. Bordoff Pascal J. NoelPay-As-You-DriveAuto Insurance:A Simple Way to ReduceDriving-Related Harmsand Increase Equity The Brookings Institution
  2. 2. The Hamilton Project seeks to advance America’s promise ofopportunity, prosperity, and growth. The Project’s economicstrategy reflects a judgment that long-term prosperity is bestachieved by making economic growth broad-based, byenhancing individual economic security, and by embracing arole for effective government in making needed publicinvestments. Our strategy—strikingly different from thetheories driving economic policy in recent years—calls for fiscaldiscipline and for increased public investment in key growth-enhancing areas. The Project will put forward innovativepolicy ideas from leading economic thinkers throughout theUnited States—ideas based on experience and evidence, notideology and doctrine—to introduce new, sometimescontroversial, policy options into the national debate withthe goal of improving our country’s economic policy.The Project is named after Alexander Hamilton, thenation’s first treasury secretary, who laid the foundationfor the modern American economy. Consistent with theguiding principles of the Project, Hamilton stood for soundfiscal policy, believed that broad-based opportunity foradvancement would drive American economic growth, andrecognized that “prudent aids and encouragements on thepart of government” are necessary to enhance and guidemarket forces. TH E HAMILTON PROJECT Advancing Opportunity, Prosperity and Growth
  3. 3. TH E HAMILTON PROJECT Advancing Opportunity, Prosperity and GrowthPay-As-You-Drive Auto Insurance: A Simple Way to Reduce Driving-Related Harms and Increase Equity Jason E. Bordoff Pascal J. Noel NOTE: This discussion paper is a proposal from the authors. As emphasized in The Hamilton Project’s original strategy paper, the Project was designed in part to provide a forum for leading thinkers across the nation to put forward innovative and potentially important economic policy ideas that share the Project’s broad goals of promoting economic growth, broad-based partici- pation in growth, and economic security. The authors are invited to express their own ideas in discussion papers, whether or not the Project’s staff or advisory council agrees with the specific proposals. This discussion paper is offered in that spirit. J U LY 2 0 0 8
  4. 4. Pay- As- Y ou-Drive A uto Insurance Abstract The current lump-sum pricing of auto insurance is inefficient and inequitable. Drivers who are similar in other respects—age, gender, location, driving safety record—pay nearly the same premiums if they drive five thousand or fifty thousand miles a year. Just as an all-you-can-eat restaurant encourages more eat­ing, all-you-can-drive insurance pricing encourages more driving. That means more accidents, congestion, carbon emissions, local pollution, and dependence on oil. This pricing system is inequitable because low-mileage drivers subsidize insurance costs for high-mileage drivers, and low-income people drive fewer miles on average. In this discussion paper, we propose and evaluate a simple alternative: pay-as-you-drive (PAYD) auto insurance. If all motorists paid for accident insurance per mile rather than in a lump sum, they would have an extra incentive to drive less. We estimate driving would decline by 8 percent nationwide, netting society the equivalent of about $50 billion to $60 billion a year by reducing driving-related harms. This driving reduction would reduce carbon dioxide emissions by 2 percent and oil consumption by about 4 percent. To put it in perspective, it would take a $1-per-gallon increase in the gasoline tax to achieve the same reduction in driving. Unlike an increase in the gas tax, PAYD would save most drivers money regardless of where they live. We estimate almost two-thirds of households would pay less for auto insurance, with each of those households saving an average of $270 per car. Despite the large social benefits from PAYD, there are currently several barriers to its widespread adoption, including the cost to monitor miles traveled and some state insur- ance regulations. In order to facilitate the spread of PAYD, we propose a three-part strat- egy. First, states should pass legislation permitting mileage-based insurance premiums. Second, the federal government should increase the funding available to PAYD pilot pro- grams by $15 million over five years. Finally, since the monitoring costs may exceed the expected benefit of PAYD to insurance firms but are much smaller than the social benefit, the federal government should offer a $100 tax credit for each new mileage-based policy that an insurance company writes, to be phased out once 5 million vehicles nationwide are covered by PAYD policies. In short, PAYD represents a win-win policy. What is good for drivers, in this case, is also good for society. Copyright © 2008 The Brookings Institution THE H AMILTON PROJECT    |   the broo king s ins titution
  5. 5. Pay-A s -Yo u - D r i v e A u t o In s u r a n c e Contents 1. Overview 5 2. The Challenge 7 3. A New Approach 12 4. A Three-Part Strategy to Encourage PAYD Adoption 21 5. Impacts, Costs, and Benefits 24 6. Why PAYD Is Preferable to Various Alternative Insurance Models 43 7. Questions and Concerns 44 8. Conclusion 48 Appendix 49 References 52 w w w.hamiltonproject. org   |    J U LY 2 0 0 8
  6. 6. Pay- As- Y ou-Drive A uto Insurance THE H AMILTON PROJECT    |   the broo king s ins titution
  7. 7. Pay-A s -Yo u - D r i v e A u t o In s u r a n c e1. OverviewI f you are like most Americans, you probably eat driving in the aggregate. It saves money for those too much when you dine at an all-you-can-eat who drive less than average, shifting the cost to buffet. Now imagine that Americans paid for those who drive more and who thus are responsiblegasoline on an “all-you-can-drive” basis—paying a for more accidents. Geography is already a key riskset fee each year for as much as they use. People factor in pricing insurance, so those in rural areaswould invariably drive more since there would be where people drive greater distances will not be dis-little cost to doing so. The idea may seem absurd, proportionately impacted because their premiumsbut that is how auto insurance is priced today. Driv- will be determined relative to how many miles theers who are similar in all respects—age, gender, average driver in their area drives.driving record—pay roughly the same premiumswhether they drive five thousand or fifty thousand With insurance costs that vary with miles driven,miles a year, even though the likelihood of being in- we estimate that drivers nationwide would reducevolved in a collision increases with each mile driven. miles traveled by about 8 percent. To put that in(Some firms do offer a modest discount for driving perspective, it would take a $1-per-gallon increasebelow a certain number of miles, but even that is in the gas tax to achieve an equivalent reduction inbased on a self-reported estimate.) Moreover, just as vehicle miles traveled (VMT). An 8 percent reduc-people consume more food when they do not bear tion in VMT would yield net social benefits of $50the cost of extra food, so too do they drive more billion to $60 billion per year, largely from reducedwhen they do not bear the cost of insurance for the congestion and accidents. It also would reduceadditional miles driven. The increased driving that carbon emissions by 2 percent and oil consump-results imposes significant costs on society: more tion by around 4 percent. In addition, PAYD cantraffic accidents, increased congestion, decreased achieve these gains while actually reducing the costair quality, growing greenhouse gas (GHG) emis- of driving for most drivers. Almost two-thirds ofsions, and deepening dependence on oil. The cur- households would enjoy reduced premiums underrent system is also inequitable, because low-mileage PAYD, and the average savings for those two-thirdsdrivers subsidize the accident costs of high-mileage of households would be $270 per car a year, equaldrivers, and low-income people drive fewer miles to 28 percent of the average annual U.S. car insur-on average. ance premium.A better approach is simple and obvious: pay-as- Despite the large social benefits from PAYD, thereyou-drive (PAYD) auto insurance. With PAYD, in- are currently several barriers to its widespreadsurance premiums would be priced per mile driven. adoption. For one, insurance companies would in-Pricing insurance per mile is more equitable be- cur costs to monitor miles traveled and develop newcause low-mileage drivers would no longer sub- pricing models whereas most of the benefits wouldsidize high-mileage drivers. With insurance costs accrue to other insurance companies and society asthat vary with miles driven, people would be able a whole. In addition, insurance regulations in manyto save money by reducing their driving, and this states prohibit or pose significant barriers to pricingincentive would lead to decreased driving-related insurance by the mile.externalities like carbon emissions and congestion.PAYD is a simple and pragmatic reform. Moreover, In order to overcome these barriers and facilitate theit is more politically feasible than alternatives like a spread of PAYD, we propose a three-part strategy.gas tax because PAYD does not increase the cost of First, states should pass legislation permitting mile- w w w.hamiltonproject.org   |    J U LY 2 0 0 8
  8. 8. Pay- As- Y ou-Drive A uto Insuranceage-based insurance premiums. Second, the federalgovernment should increase the funding availableto PAYD pilot programs to $15 million over fiveyears. Finally, the federal government should offera $100 tax credit for each new mileage-based policythat an insurance company writes, to be phased outonce 5 million vehicles nationwide are covered byPAYD policies. THE H AMILTON PROJECT    |   the broo king s ins titution
  9. 9. Pay-A s -Yo u - D r i v e A u t o In s u r a n c e2. The ChallengeE ver since Henry Ford made cars affordable to erably to bring home to the automobile user the the masses, Americans have had a love affair costs he imposes in a manner that will appropriately with their automobiles. We run our cars almost influence his decisions” (Vickrey 1968).3 trillion miles a year, using 140 billion gallons of gaso-line in the process. Currently, car owners pay roughly Notably, the relationship between VMT and acci-the same in automobile insurance regardless of how dents is not proportional. In aggregate, motoristsmany miles they drive. To be sure, some companies that drive more tend to have fewer accidents peroffer a discount if a driver drives below a certain num- mile. One who drives thirty thousand miles a yearber of miles. For example, Liberty Mutual and State will be involved in fewer than three collisions perFarm offer a discount of about 15 percent if a driver million miles driven, whereas one who drives fewerdrives less than 7,500 miles annually, but even that than eight thousand miles a year will be involved ondiscount is based on a self-reported estimate, which average in more than seven collisions per millionlimits the amount of discount offered.1 There is no miles driven (Figure 2).reliable method to verify the number of miles drivenor adjust premiums accordingly. Progressive, one of Some reasons the relationship is not proportionalthe nation’s largest auto insurers, explains as follows: include the following (Litman 2005):“A principal problem with . . . conventional insur-ance determination systems is that much of the data • Higher-mileage drivers tend to be more skilledgathered from the applicant is not verifiable, and even than lower-mileage drivers.existing public records contain only minimal infor- • Newer (thus, mechanically safer) vehicles tendmation, much of which has little relevance towards an to be driven more.assessment of the likelihood of a claim subsequently • Urban drivers tend to have higher crash ratesoccurring.”2 and lower annual mileage. • High-mileage drivers tend to do a greater shareWith this pricing scheme, there is almost no mar- of driving on safer, grade-separated highways.ginal insurance cost to driving another mile, yet thelikelihood that a driver will be involved in a collision While a comparison across different vehicles doesnecessarily increases with mileage driven.3 For exam- not yield a proportional relationship, it is impor-ple, as seen in Figure 1, a vehicle that travels twenty tant to bear in mind that the risk of any individualthousand miles is roughly twice as likely to have an vehicle being involved in an accident necessarilyaccident as one that travels less than five thousand declines roughly proportionally with a reductionmiles. As Nobel Prize–winning economist William in VMT. Thus, while a high-mileage motorist mayVickrey puts it, “The manner in which [auto insur- be only twice as likely to have an accident as a low-ance] premiums are computed and paid fails mis- mileage motorist who drives one-fourth as much,1. Liberty Mutual offers a discount in forty-five states plus the District of Columbia of between 5 and 18 percent to drivers who drive less than 7,500 miles per year. Most of these states offer a 13 percent discount. State Farm considers those who drive less than 7,500 miles a year to be low-annual-mileage drivers, and they receive a 15 percent discount on average. Geico also offers a discount to those who drive less than 7,500; the average amount of the discount was unavailable to us, however.2. U.S. Patent No. 6,868,386 (filed May 15, 2000).3. Of course, insurance premiums are not completely outside a driver’s control: they do in some sense respond to mileage. Someone consid- ering whether to drive another mile knows there is a chance she may get into an accident or be cited for a moving violation and thereby cause her insurance rates to jump in subsequent years. But given that the average driver has an accident every twenty years or two hundred thousand miles or so, this incentive does little on the margin (Insurance Information Institute n.d.). w w w.hamiltonproject.org   |    J U LY 2 0 0 8
  10. 10. Pay- As- Y ou-Drive A uto InsuranceFigure 1Yearly Accident Claims by Annual Mileage 0.10 Collision Claims 0.09 Property Damage Claims 0.08 Bodily Injury ClaimsAccident Claims per Year 0.07 0.06 0.05 0.04 0.03 0.02 0.01 0 0 5,000 10,000 15,000 20,000 25,000 30,000 Annual Mileage ClassSource: Progressive 2005.if the high-mileage driver reduces her VMT by 10 risk information as proxies. Their current pricingpercent she will also reduce her risk of getting into structure is thus already picking up some of thean accident by about 10 percent. That relationship mileage differences between drivers. Companiesis likely to be proportional because the individual may indirectly capture mileage risk, for example, bydriver’s attributes do not change if she drives less. charging drivers who say they live far from work andReliable data on individual VMT variations and ac- who have poor accident histories a higher premiumcidents to confirm this proportional relationship than they charge those with the opposite character-for individual drivers are lacking because insurance istics. This does not mean actual mileage should notcompanies generally do not make accident data be taken into account, however. Pricing per milepublicly available. Unlike snapshot comparisons of would still be a far more accurate way to price basedvehicles across the distribution of VMT (such as on risk than using such proxies, particularly givenin Figure 1), however, a better way to answer this that accidents necessarily and evidently (Figure 1)question is to look at accident decline overall when increase with mileage. A motorist’s driving recordthere is an aggregate drop in VMT for all vehicles also is not an adequate proxy since the likelihooddue to exogenous factors. For example, a recession of having an accident is so low; neither a low-mile-in 1981–82 caused a 10 percent reduction in VMT age driver nor a high-mileage driver may thus haveand a 12 percent reduction in insurance claims in been in an accident, even though the likelihood ofBritish Columbia (Litman 2005), which is a more- the high-mileage driver being in one is still severalthan-proportional decline. times higher. Indeed, in interviews we conducted with insurance companies and their actuaries, thereWhile insurance companies do not price auto in- was consistent agreement that charging for mile-surance based on miles driven, they do use other age would be preferable to the current system. One THE H AMILTON PROJECT    |   the broo king s ins titution
  11. 11. Pay-A s -Yo u - D r i v e A u t o In s u r a n c e Figure 2 Accident Claims per Million Miles by Annual Mileage 9 Collision Claims 8 Property Damage ClaimsAccident Claims per One Million Miles Bodily Injury Claims 7 6 5 4 3 2 1 0 0 5,000 10,000 15,000 20,000 25,000 30,000 Annual Mileage Class Source: Authors’ calculation based on Progressive 2005. company executive called charging per mile the remain relatively small. The result is that with the pos- “last barrier of insurance pricing.” sible exceptions of the decisions as to whether to drive to work or use public transportation, and of the decision Fundamentally, the pricing structure of auto insur- as to whether younger members of the family are al- ance has not changed since Vickrey’s observations lowed to drive at all, the added exposure to risk involved forty years ago. In 1968, he wrote the following: in added usage is not brought to bear on the decision. (Vickrey 1968, 470–71) The basic difficulty is that the insurance premium ap- There are two problems with the current lump-sum pears to the individual automobile owner almost entire- pricing structure. First, low-mileage drivers end up ly as part of the fixed cost of owning a car. The amount subsidizing high-mileage drivers in each risk class, of the premium, given the coverage he selects, is fixed even though the former are responsible for fewer by factors largely independent of most of the decisions accidents. This problem is particularly disturbing that are at all marginal as to how much he will use his given that low-income people tend to drive less on car. The only attempts that are made to vary premiums average (Figure 3). In 2001, households with in- in relation to use are typically to classify the risk accord- comes above $100,000 drove each of their vehicles ing to whether and how far the car is driven to work or on average 25 percent more miles than households whether it is used for business; the classifications are with incomes below $25,000. Recent survey data very broad and to a considerable extent are based on from the Greater Vancouver Regional District show the unverified statements of the applicant. Moreover, an even stronger correlation (Litman 2007b). the variations in premiums based on such classifications w w w.hamiltonproject.org   |    J U LY 2 0 0 8
  12. 12. Pay- As- Y ou-Drive A uto Insurance Second, lump-sum pricing leads to an inefficiently Vickrey (1968) explained, is an increased number of high level of driving and thus makes drivers worse VMT, and all these extra miles also impose signifi- off. Since insurance is priced in a lump-sum fashion, cant costs on society. Burning fossil fuels in vehicles a driver does not take the marginal cost of insur- releases carbon dioxide (CO2), a heat-trapping gas ance into consideration when deciding how much that can remain in the atmosphere for more than a to drive (the way she might consider the costs of hundred years. CO2 is the principal GHG respon- fuel and maintenance, for example). The result is sible for global warming, a human-caused process that some of the miles she drives are not worth the that is now widely accepted as real. This warming marginal cost to her if she were actually to pay for may cause massive climatic shifts, characterized insurance per mile driven. If she could pay for in- by changes in precipitation and runoff, increased surance by the mile and save that cost by reduc- flooding, drought, and more-frequent and severe ing miles driven, she would do so and be better off storms. Continuing to burn such high quantities of by paying lower total premiums. PAYD makes this fossil fuels could trigger climate change costing the possible. A driver only needs accident-related car economy up to 1.5 percent of GDP (Cazorla and insurance when she is driving. But unlike for almost Toman 2000). every other good available to consumers, under the current system a driver cannot pay less for car in- Increased gasoline use also deepens our nation’s surance by consuming less of it. reliance on oil, increasingly understood as a threat to economic and national security. Oil price shocks The result of this inefficient pricing structure, as have played a role in nine of the ten U.S. reces- Figure 3 Average Mileage per Vehicle, by Household Income Level 15,000 14,000Average Annual Mileage 13,000 12,000 11,000 10,000 9,000 8,000 0 0 00 00 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 00 50 50 50 50 50 50 50 50 50 50 50 50 50 50 00 ,5 ,5 ,0 2. 7, 2, 7, 2, 7, 2, 7, 2, 7, 2, 7, 2, 7, 0, $2 $7 00 $1 $1 $2 $2 $3 $3 $4 $4 $5 $5 $6 $6 $7 $7 $9 $1 Annual Household Income Source: Authors’ calculation based on data from the 2001 National Household Transportation Survey. 10 THE H AMILTON PROJECT    |   the broo king s ins titution
  13. 13. Pay-A s -Yo u - D r i v e A u t o In s u r a n c esions since World War II. Oil wealth strengthens cident because there is one more car with whichoil-exporting authoritarian governments and limits another car could potentially collide. Thus, evenAmerica’s foreign policy options for meeting their adding the safest of drivers to the road increasesgrowing threats. Reducing oil consumption is the total accident costs because there are more cars onkey factor in improving America’s energy security. the road. Economist Aaron Edlin (2003) finds that the elasticity of accidents with respect to VMT mayEvery additional mile driven also adds gases to the be around 1.7, meaning that a 10 percent reductionatmosphere that contaminate the air we breathe. Ni- in VMT would lead to a 17 percent reduction in to-trogen oxides (NOX) and hydrocarbons (HC) have tal crashes. The costs of accidents—in terms of livesserious and negative impacts on human health. lost, injuries to victims, lost productivity, property damages, medical costs, travel delays, and legal ex-Even greater than these environmental and secu- penses—are staggering. Ian Parry, Margaret Walls,rity concerns is the estimated impact of increased and Winston Harrington (2007) estimate that acci-driving on traffic congestion. Urban traffic conges- dents cost the United States $433 billion in 2000.tion in 2005 caused the average peak-period trav-eler to spend thirty-eight extra hours in travel time To be clear, PAYD would not address these driv-(Schrank and Lomax 2007). The number of urban ing-related externalities in full by forcing driversareas with more than forty hours of annual delay to internalize the cost of the social harm caused byper peak traveler has grown from only one in 1982 their activities. Doing so would require a set of op-to twenty-eight in 2005. This wasted time costs the timized user fees specifically calibrated to capturecountry $78 billion annually, which is almost exact- each externality. By reforming the way auto insur-ly the size of the entire U.S. federal transportation ance is priced, however, PAYD also begins to makebudget. progress on all these issues by creating an incentive to reduce driving. PAYD is a politically viable wayFinally, more driving increases the number of auto to make progress on all these issues because it actu-accidents. Even when a driver is not at fault, merely ally reduces the cost of driving for most people.being on the road increases the likelihood of an ac- w w w.hamiltonproject.org   |    J U LY 2 0 0 8 11
  14. 14. Pay- As- Y ou-Drive A uto Insurance3. A New ApproachA relatively simple and pragmatic way to ad- prehensive coverage, which covers damage caused dress the problems with the way auto in- by fire, theft, vandalism, and weather. The majority surance is priced also would help reduce of car insurance premiums cover risks that vary di-all these driving-related externalities. Under PAYD rectly with mileage; these premiums thus would beauto insurance, the price of auto insurance would switched to a per-mile rate. These include liabilitybe tied to the number of miles driven. Other rat- coverage (for the costs of bodily injury and prop-ing factors such as location, age, vehicle type, and erty damage imposed on others when the insureddriving record still would be incorporated into this causes an accident); collision or accident coverageprice, so higher-risk drivers would pay more per (for medical payments, personal injury costs, andmile than lower-risk drivers. PAYD ensures that collision damages incurred by the insured and herlow-mileage drivers stop subsidizing the accident vehicle when the insured causes an accident); andcosts of high-mileage drivers. By allowing people to uninsured motorist coverage (for damages to thesave money by driving less, PAYD creates incentives insured and her vehicle caused by an uninsuredfor reducing the various costs that driving imposes motorist). According to personal email communi-on society. cation with Progressive County Mutual Insurance Company, liability, accident, and uninsured cover-A small portion of current premiums cover risks not age make up 89 percent of the typical premium itdirectly related to mileage, so they would remain collects (Figure 4).unchanged. This includes, most prominently, com-Figure 4Distribution of Progressive Auto Insurance Premium, by Type of Coverage Other 2% Bodily Injury 24% Comprehensive 9% Uninsured/Underinsured Motorist 5% Medical Payments 2% Property Damage 22% Personal Injury Protection 8% Collision 28%Source: Personal e-mail communication with Progressive, 5/14/08.12 THE HAMILTON PROJECT    |   the broo king s ins titution
  15. 15. Pay-A s -Yo u - D r i v e A u t o In s u r a n c eA. Designing PAYD Insurance various publicly available odometer readings into its vehicle history reports, available to insuranceThere are several specific design features firms will companies on any car and light truck model 1981need to choose. Under the simplest model, motor- or later. Individual consumers can have unlimitedists would prepay for the miles they expect to drive access to vehicle history reports for one month forin a year and then pay more or receive a refund at the $29.99, but commercial rates vary by volume andend of the year depending on actual miles driven. In purpose.order to avoid people purposely underestimating todefer payments, firms might automatically set each Another option is to use electronic devices that au-year’s estimate at the total number of miles driven tomatically record and transmit mileage data. Mostin the prior year. Drivers would be free to change new cars already electronically record mileage inthat estimate if they anticipated driving more or less the engine computer, and new technologies suchthan the prior year, but might pay a penalty if the as global positioning system (GPS) transpondersactual miles driven exceeded the estimate by more provide a means of wireless transmission. Table 1than a certain amount (say, 20 percent). Alternative- describes several technology options available toly, insurance companies could bill motorists based insurers. Davis, an American company based inon their monthly or bimonthly vehicle mileage, just California, offers its CarChipPro technology toas utilities currently do. This would require more- individual consumers for $119. The CarChipProfrequent mileage data collection, however, probably is a small data-collection device that plugs in tovia a telematic device in the vehicle. Though more a car’s on-board diagnostics (OBD) port, readingcostly, this approach may result in larger social ben- and storing data from the car’s on-board comput-efits by effecting a larger reduction in VMT, as dis- ers. (OBD ports have been mandatory for new carscussed further below. in all states since 1996.) Consumers connect the small device to their home computers via a USBPAYD requires verified mileage data. One way to cable to download driving data. A wireless versiondo that is through odometer audits. Licensed pro- is available, tailored to fleet managers, which allowsfessionals such as safety and emissions inspectors for wireless transmission of the data when a vehi-could verify mileage every year, or even more fre- cle is parked near a receiving transponder. Otherquently. They would have to ensure the odometer versions of similar technology are available fromwas not tampered with and then transmit the data companies such as IMS and Sky-meter. Both theseto insurance companies. Drivers would bear the firms market specifically to insurance companiescost of taking their cars in to be checked, though and offer wireless data transmission. The devicein some states and many urban areas annual inspec- from Sky-meter is leased to the consumer (or com-tions are already required at which an odometer pany) for a $5 monthly fee plus a portion of thereading could be done (in some cases this is being insurance premium. Conversations with auto in-done now). Currently, nine states have mandatory surance executives and manufacturers reveal pricesannual inspections for all vehicles, four states have have been declining and are expected to continuemandatory inspections for all vehicles every two declining significantly as the technology advances.years, and various localities in nine other states have Most of these devices do not record or transmit in-mandatory inspections every one or two years. One formation about a driver’s location, and even thosecompany, CARFAX, claims to aggregate all these devices that have such a capacity may be tailored to eliminate any privacy concerns. w w w.hamiltonproject.org   |    J U LY 2 0 0 8 13
  16. 16. Pay- As- Y ou-Drive A uto InsuranceTable 1Cost of Commercially Available Telematic Mileage-Recording Devices Manufacturer Data recorded: Method of transmission Installation cost Monthly/yearly fee Distance, speed, time Davis CarChipPro Distance, speed, time USB cable/port $119 None (could include other (customer loaded) features for higher price ) CarChipFleetPro Distance, speed, time USB cable/port $169 (plus a $395 None (could include (customer loaded) charge for software, other features) one per fleet) Can also be used wirelessly with a $200 base unitIMS iPAID Distance, speed, time USB cable (manual $60–$80 for Varies upload on PC) or mileage Bluetooth (automatic measurement only upload) or cellular (GPRS) Sky-meter Distance, speed, time GPRS/CDMA $50 - $250 $5 per month plus (incl. other features) (other protocols activation fee 5%–8% of monthly available at extra premium charge) (depending on volume)OnStar Distance, speed, Automatic through GPS First year free for $18.95 per month time, (incl. other new GM cars after one year features) (only available for GM) Source: Company Web sites and personal communications with authors.14 THE H AMILTON PROJECT    |   the broo king s ins titution
  17. 17. Pay-A s -Yo u - D r i v e A u t o In s u r a n c eOther models of monitoring mileage exist. For than fifteen thousand miles: 1–2,500 miles (50 per-example, one is being developed by start-up insur- cent discount), 2,501–5,000 (33 percent), 5,001–ance firm MileMeter. Under its model, individuals 7,500 (28 percent), 7,501–10,000 (20 percent),would go online and purchase a specific number of 10,001–12,500 (11 percent), and 12,501–15,000 (5miles of coverage. A driver would not be covered percent). GMAC currently has about twenty-fivein the event of an accident if the car’s odometer in- thousand OnStar subscribers, some of whom aredicated that the driver had exceeded the amount signed up for the mileage discount. OnStar is freeof insurance purchased. This approach negates the for the first year for new GM car owners, but costsneed for odometer audits, though it may raise con- $18.95 per month subsequently.cerns about uninsured cars. MileMeter would stillcollect odometer readings from vehicle emissions, While the precise manner in which PAYD insur-maintenance, and registration databases (similar to ance is implemented may vary, how the pricing isthe information available from CARFAX). designed can have a significant impact on the extent to which PAYD reduces VMT. Insights from behav-Some established companies are already using ioral economics about loss aversion, for example,monitoring technology to offer mileage discounts teach us that consumers would prefer to avoid aon insurance premiums. In June 2008, Progressive surcharge than to receive a discount (Kahneman,announced a national rollout of its MyRate insur- Knetsch, and Thaler 1986). PAYD would thus beance program. Under MyRate, cars driven less of- more likely to result in a greater reduction in VMTten, in less-risky ways, and at less-risky times of day if drivers were charged an insurance price for eachcan receive a lower premium (Progressive 2008). mile driven than if they were to receive discountsAccording to Progressive, the impact on premiums for driving fewer miles than expected.could be anywhere from a 60 percent discount to a9 percent surcharge. The MyRate program evolved In addition, the frequency with which the cost of in-out of Progressive’s TripSense program in Michi- surance is communicated to drivers and the claritygan, Minnesota, and Oregon. Under that program, and simplicity with which that cost is communicatedparticipating drivers receive a TripSensor based on will affect how much drivers change their behavior.the Davis CarChip technology. The TripSensor re- To be clear, the goal of PAYD pricing should not becords how much, how fast, and when the vehicle is to maximize driving reductions at all cost, but ratherdriven—information that is used to calculate dis- to send accurate and consistent price signals to con-counts of 5–20 percent when the customer renews sumers, enabling them to make individually opti-the policy. (This program is thus not true per-mile mal decisions. Economist Amy Finkelstein recentlypricing.) The collected data and the potential dis- found that when tolls switch from cash payments tocount are communicated to the customer before electronic toll collection, toll prices increase morethey are shared with Progressive. The customer has sharply. She also finds that the short-run elasticitythe option of sharing the information with Progres- of driving with respect to the actual toll declinessive and earning the discount or withholding the after electronic tolls are introduced. After rejectingdata and paying the normal premium. various other possible explanations, she concludes the most likely reason is that it is easier for toll col-General Motors Acceptance Corporation (GMAC) lectors to raise prices when drivers are less aware ofInsurance has offered mileage-based discounts to the price they are paying (Finkelstein 2007). ThisOnStar subscribers located in certain states; driv- phenomenon also can be seen in recent anecdotalers in thirty-four states are currently eligible. The evidence that hybrid car drivers alter their drivingsystem reports the odometer reading at the begin- behavior to maximize fuel economy in response toning and end of the policy term, and the customer real-time information about how many miles theyreceives discounts on a sliding scale for driving less are getting per gallon, as Michael S. Rosenwald w w w.hamiltonproject.org   |    J U LY 2 0 0 8 15
  18. 18. Pay- As- Y ou-Drive A uto Insurancepoints out in the Washington Post (“For Hybrid ny. These monitoring costs are borne by firms andDrivers, Every Trip Is a Race for Fuel Efficiency,” their customers, but the benefits spill over to otherMay 26, 2008). The lesson for PAYD would be that insurance companies, other drivers, and society asa driver is more likely to reduce VMT if she re- a whole. If an insurance company is able to reduceceives frequent price signals about the cost of her the driving of its insureds, substantial savings willinsurance than if she receives an insurance bill at accrue to other insurance companies too, insofar asthe end of the year based on annual VMT, such as their insureds are less likely to be involved in ac-after an annual odometer reading. cidents if fewer vehicles are on the road. Reduced driving benefits all other drivers also, via reducedB. Why PAYD Has Not Spread More congestion. Finally, savings will accrue to society asWidely on its Own a whole from reductions in local and global pollu- tion and oil dependence.If PAYD is more efficient and equitable, why is itnot already offered? There are three market and Despite the substantial social savings from switch-regulatory failures that prevent PAYD from emerg- ing to per-mile auto insurance, there may be verying on its own: monitoring costs, state insurance little direct savings that actually accrue to insuranceregulations, and patents. companies. We calculate in §4 that if insurance com- panies captured all the benefit from reduced acci-i. Monitoring costs. dents of their insureds they would save $34 per yearFirst, the monitoring costs borne by insurers may per vehicle that switched from traditional lump-sumexceed the private benefits to firms, even though premiums to PAYD. We also calculate that the aver-the private benefits are far less than the potential age low-mileage driver might save as much as $318social benefits (see Edlin 2003; Rea 1992; William- per vehicle as PAYD eliminates the implicit subsidyson, Olson, and Ralston 1967). In order to price in- to high-mileage drivers. Those savings would besurance per mile, insurance firms or their customers offset in aggregate by higher premiums charged towould need to incur the cost of verifying mileage, high-mileage drivers, but in the short term it mighteither through odometer checks (fitted with tam- be possible for first-mover auto insurers to captureper proof technology) or through a device fitted in some of these gains by signing up low-mileage driv-each vehicle. Though in theory odometer readings ers currently with other insurers. In the long run,could be inexpensive procedures if done on a wide- any private gain would likely be much lower sincespread basis, there currently is no infrastructure of expected benefits would be competed away oncecertified providers to provide reliable odometer other firms also adopt PAYD. Moreover, it may bereadings that insurance firms can use. Without the difficult for insurance firms to capture fully theseability to rely on the accuracy of a service station’s gains from any particular customer, given the highodometer reading, firms would have to employ their churn rate for customers in the auto industry. Toown teams of odometer readers, which might be capture that return from new or existing customerscost prohibitive. They also would have to compen- the firm must develop a new billing and administra-sate customers somehow for their time if they were tive infrastructure, retool their advertising to edu-not able to combine the odometer reading with cate consumers, and develop new actuarial modelsnormal inspections or maintenance work. Install- to determine how to price per-mile premiums, alling GPS or similar technology to monitor VMT of which would add to the cost.might be expensive. As seen in Table 1, it can costwell over $100 just to install a device that records The private incentive pales in comparison to theVMT, and can cost several times that amount, in- social benefit from switching a vehicle to PAYD,cluding monthly service fees, for a device that also which we calculate in §4 to be $257 per vehicle.wirelessly transmits the information to the compa- In other words, there are large positive externali-16 THE HAMILTON PROJECT    |   the broo king s ins titution
  19. 19. Pay-A s -Yo u - D r i v e A u t o In s u r a n c eties arising from PAYD that cannot be captured by A 2002 survey by the Georgia Institute of Technol-private firms offering it. The significant discrepancy ogy of forty-three state insurance commissionersbetween the social and private benefits suggests that found that 37 percent indicated state regulationseven if the benefits to the firm and its insureds do would not permit PAYD insurance in their statenot justify an insurance company’s incurring the (Guensler, Amekudzi, Williams, Mergelsberg, andmonitoring and plan development costs, the full Ogle 2002).4 Even in states in which regulationsocial benefits would justify the costs. does not explicitly prohibit PAYD, certain legisla- tive reforms might be needed because of potentialii. State insurance regulations. conflicts with existing law. Michigan, for example,Second, state insurance regulations can pose a barri- requires an upfront statement of the premiumer to PAYD. State regulators must explicitly approve charge, and the policy must have an expiration datethe type of insurance policies that insurers can offer; and must be renewable. This might be difficult un-in several states, current regulations appear to pro- der PAYD since the total premium for any periodhibit pricing insurance per mile. Current efforts to is unknown at the start of the period (although apermit PAYD pricing in California are indicative of program could theoretically be designed to chargethe types of barriers that PAYD faces due to various a fixed up-front cost and then offer per-mile dis-state laws and regulations. Proposition 103 requires counts based on miles actually traveled). In Georgia,that automobile insurance premiums be based pri- companies need to ensure that their policy requiresmarily on three factors in descending order of im- a down payment for at least sixty days of coverageportance: (1) driving safety record, (2) annual miles and that the minimum insurance term was not lessdriven, and (3) years of driving experience. Despite than six months, but such temporal requirementsthis law’s intention to ensure that mileage be taken might be inconsistent with certain per-mile pricinginto strong consideration, the regulations that imple- models. Until the law was modified in 2001, thement it may actually stand in the way of offering true Texas Transportation Code required proof of insur-PAYD pricing. Insurance firms in California typically ance (for vehicle title changes) to be a one-monthoffer only very wide mileage bands because they must policy, which a PAYD policy would not satisfy;accept self-reported estimates of miles driven. As a presumably several other states may have a similarresult, low-mileage drivers are paying too much for regulatory obstacle to PAYD.auto insurance and are subsidizing high-mileage driv-ers. The regulations implementing Proposition 103, Even in states that permit PAYD in theory, severalhowever, prohibit an insurance firm from charging insurance commissioners indicated that regulationsa PAYD customer whose VMT was verified a lower in their state would prohibit certain commonlypremium than a customer of identical risk profile proposed methods of pricing PAYD. Roughly halfwhose VMT was not. These implementing regula- of the states responding would not permit a PAYDtions thus preclude offering an insurance product pricing model whereby premiums would be basedto low-mileage drivers that more accurately reflects on the driver’s current annual fees and reportedtheir accident risk because other low-mile drivers in mileage (Guensler et al. 2002). The survey respon-traditional insurance plans would be paying a differ- dent in Tennessee, for example, indicated the stateent (and higher) rate. Currently, the California De- would not accept such a pricing method because thepartment of Insurance has undertaken a rulemaking structure constituted a retrospective rating scheme,process, and a bill is pending in the state Senate (AB which its regulations do not allow. In West Virginia,2800), both of which are aimed at overcoming such the regulations require that customers be insured atobstacles to offering PAYD in California. all times, and PAYD might mean that drivers would4. This survey is several years old and was imprecise in its design, thus its findings should be viewed with considerable caution. Nonetheless, it is instructive of the types of regulatory barriers PAYD faces. w w w.hamiltonproject.org   |    J U LY 2 0 0 8 17
  20. 20. Pay- As- Y ou-Drive A uto Insurancelose coverage if they exceeded the prepaid number such competitive forces are blunted in the insur-of miles (although an insurer could structure a ance sector, however, innovations like PAYD mayPAYD program to ensure that participating drivers be discouraged.are never uninsured). iii. Patents.Furthermore, auto insurance agencies must have Finally, for the past decade Progressive has aggres-their rate plans approved by state regulators each sively sought patents around innovations in tele-year. Whereas regulators typically approve as a rou- matic auto insurance (Tom Bakos Consulting andtine matter current rate plans submitted for reap- Markets, Patents and Alliances 2004).5 Indeed, evenproval, they closely scrutinize changes to rate plans. the title “Pay As You Drive” is registered to Progres-In interviews, insurance firms argue that they are sive, which precluded Unigard from using that titlethus reluctant to attract extra scrutiny from regu- for its per-mile insurance product launched as partlators by offering innovative products. Insurance of a federally funded pilot program in King County,regulators whose primary charge is to ensure that Washington. In interviews with the authors, severalinsurance consumers are not harmed may be reluc- auto insurance executives have identified these pat-tant to approve a rate plan as revolutionary as PAYD, ents as a barrier to adoption of PAYD insurance.given such uncertainty about what the ultimate rateplans will look like. Regulators’ reticence is likely Progressive holds four patents related to PAYD,exacerbated by the limited familiarity with PAYD each of which is quite broad in scope. Three of theseand widespread misunderstandings regarding its ef- patents involve a “monitoring system for determin-fects. For example, the New York respondent to the ing and communicating a cost of insurance.”6 A fewaforementioned survey stated that cents-per-mile of the most relevant claims include the following:programs would not be equitable because upstatedrivers tend to drive more than downstate drivers • method of determining a cost of automobile “Ado (Guensler et al. 2002). (In practice, premiums are insurance for a selected period based upon moni-risk-adjusted, so upstate drivers would be charged toring, recording and communicating data rep-more only if they drive more than the average simi- resentative of operator and vehicle driving char-larly situated driver.) acteristics during said period, whereby the cost is adjustable by relating the driving characteristicsFinally, the heavily regulated insurance field may to predetermined safety standards.”7pose barriers for new entrants, a concern confirmedby conversations with start-up firms in the industry. • process for acquiring and recording vehicle “AEstablished firms may be content with the status insurance related data during a time period viaquo, notwithstanding its inequities and failings. an on-board computer and recording system forInertia, high start-up costs, and uncertainty may adjusting an insurance cost during the time pe-discourage them from adopting a radically new riod.”8business model like PAYD. In well-functioningcompetitive markets, new entrants might promote • method of insuring a vehicle operator for a se- “Asuch innovation at a more rapid pace. To the extent5. MileMeter also holds patents related to PAYD. We focus on Progressive’s, however, since they are broadest in scope.6. U.S. Patent No. 6,868,386 (filed May 15, 2000) is a continuation-in-part of U.S. Patent No. 6,064,970 (filed August 17, 1998), which is a continuation of U.S. Patent No. 5,797,134 (filed January 29, 1996). A continuation is a copy of an original patent application except the inventor adds new “claims,” and a continuation-in-part is similar except that the inventor adds a description of the improvements to an invention and submits claims covering the improvements.7. U.S. Patent No. 5,797,134 (filed January 29, 1996) (claim 1).8. U.S. Patent No. 5,797,134 (filed January 29, 1996) (claim 12).18 THE H AMILTON PROJECT    |   the broo king s ins titution
  21. 21. Pay-A s -Yo u - D r i v e A u t o In s u r a n c e lected period based upon operator driving char- PAYD, although it is unclear to whom and on what acteristics during the period.”9 terms. Second, there might exist several ways to de- sign a PAYD system that does not infringe on Pro-• system for Internet on-line communicating “A gressive’s patents, an opinion expressed by several between an insurer and insured, of detected oper- industry and patent experts. Indeed, several firms, ating characteristics of a unit of risk for a selected such as GMAC, currently offer limited forms of period, as decided by the insurer in consideration mileage-based pricing without eliciting allegations of the detected operating characteristics.”10 of patent violation. Finally, there is a legitimate question about the extent to which Progressive’sAdditionally, Progressive holds a fourth patent, U.S. patents would be upheld in a patent infringementPatent 7,124,088 (filed July 30, 1999), which claims lawsuit, particularly in light of recent legal develop-“[a]n on-line insurance policy service system” that ments (Greenberg 2007). Of course, for any rivalcomprises various specific design elements (claim 1). firm the legal process of challenging these patents’ validity involves some risk and considerable ex-Any insurance firm interested in offering per-mile pense, and thus acts as a significant deterrent itself,insurance pricing understandably might be con- regardless of the eventual outcome.cerned that doing so would open it up to a possiblepatent infringement suit. Even if the firm believed it Regarding the validity of Progressive’s patents, ahad developed a method of per-mile insurance pric- challenge could be brought in theory on “obvious-ing that did not violate the patent or that the patent ness” grounds. To qualify for a patent, an inventionwas invalid, it might be deterred by the mere threat must not have been “obvious at the time the inven-of a lawsuit from Progressive. Defending a patent tion was made to a person having ordinary skill ininfringement suit can be enormously expensive the art to which said subject matter pertains” (35(Jaffe and Lerner 2004). As several recent studies U.S.C. §103(a) (2000)). Recently, the Supremehave lamented, the sharp increase in the number of Court clarified the meaning of “obviousness,” en-patents granted in the past two decades (including couraging lower courts to be more open to the pos-so-called bad patents that are overbroad or should sibility that an issued patent might still in fact benot have been granted in the first place) has cre- invalid for obviousness.11ated high costs for firms and individuals that need tolitigate patent lawsuits or pay unwarranted licens- As a result, the Progressive patents may be vulner-ing fees, thus deterring the pursuit of pursuing new able. Four decades ago, Vickrey (1968) observedventures (Jaffe and Lerner 2004; Lichtman 2006; that optimal auto insurance pricing should take ac-Merrill, Levin, and Myers 2004). count of VMT. “There is no real conceptual diffi- culty in charging an insurance premium accordingNonetheless, Progressive’s PAYD patents may not to mileage,” he explained. “The problem is one ofbe a serious barrier to PAYD adoption for three implementation” (Vickrey, 1968). There was thusreasons. First, Progressive has expressed (in an in- a known problem with auto insurance pricing, andterview with the authors) a willingness to license “[u]nder the correct analysis, any need or problem9. U.S. Patent No. 6,064,970 (filed August 17, 1998), (claim 4).10. U.S. Patent No. 6,868,386 (filed May 15, 2000), (claim 10).11. KSR Int’l Co. v. Teleflex Inc., 127 S. Ct. 1727 (2007). The issue in that case was a patent describing the combination of an adjustable pedal assembly with a pedal position sensor attached to the supporting shaft of the pedal assembly. Originally, adjustable pedal systems were designed to work in vehicles without computer-controlled engines. In the mid-1990s, however, the auto industry largely switched to computer-controlled engines that required electronic throttle controls. While the older systems relied on cables to link the pedal to the throttle, these new systems required the use of pedal position sensors to achieve the same interaction. KSR argued that Teleflex’s patent was invalid because the combination of an adjustable pedal assembly and a pedal position sensor was “obvious.” w w w.hamiltonproject.org   |    J U LY 2 0 0 8 19
  22. 22. Pay- As- Y ou-Drive A uto Insuranceknown in the field of endeavor . . . can provide a devices in vehicles to track VMT. (A few additionalreason for combining the elements in the manner ideas may be found here and there, such as Vickrey’sclaimed” (KSR, 127 S. Ct. at 1742). As GPS and theoretical suggestion to add the cost of insuranceother telematic systems became broadly available to the price of tires.) When there is an identifiedin automobiles, combining these devices with auto problem of the kind Vickrey observed, “there areinsurance to address the problem Vickrey identi- a finite number of identified, predictable solutions,fied may well be considered to have been obvious. [so] a person of ordinary skill has good reason toAlthough GPS-type technology was often used for pursue the known options within his or her techni-other purposes, such as tracking a vehicle’s loca- cal grasp.” If the result is success in solving the prob-tion, “familiar items may have obvious uses beyond lem, “is it likely the product not of innovation but oftheir primary purposes, and in many cases a person ordinary skill and common sense. In that instanceof ordinary skill will be able to fit the teachings of the fact that a combination was obvious to try mightmultiple patents together like pieces of a puzzle” show that it was obvious under section 103” (KSR,(KSR, 127 S. Ct. at 1742). As the Court explained, 127 S. Ct. at 1742).“[a] person of ordinary skill is also a person of or-dinary creativity, not an automaton” (KSR, 127 S. Like most legal issues, of course, there also are strongCt. at 1742). Moreover, the literature about how arguments on the other side as well. A full discus-to price auto insurance more accurately to reflect sion of all these arguments is beyond the scope ofdriver risk is replete with references to a limited this paper. The purpose of the above discussion isnumber of conceivable options: require manual simply to indicate that a rival firm wishing to offerodometer checks, add the cost of insurance to the a PAYD product may still be able to do so despiteprice of gasoline purchased at the pump, or install existing patents.20 THE HAMILTON PROJECT    |   the broo king s ins titution
  23. 23. Pay-A s -Yo u - D r i v e A u t o In s u r a n c e4. A Three-Part Strategy to Encourage PAYD AdoptionI t is hard to tell how significant each of the three by conditioning the receipt of federal discretion- barriers is by itself. The paucity of mileage-re- ary grants and formula allocations on the reform of lated insurance offerings and the nonexistence of state insurance regulations to permit PAYD.true PAYD pricing, however, indicate the barrierstogether pose a real roadblock. Given the potential Second, expand funding for PAYD pilotfor significant social benefits from wide adoption programs. There is a lack of knowledge on the partof PAYD combined with the small expected private of insurance firms and state regulators about howbenefits for firms there is a clear rationale for ef- to price and design PAYD, significant start-up costsfective public policy solutions. Policymakers should involved with being a first mover, and barriers totake the following three steps in response to these potential entrants. Given the small private benefitbarriers to encourage PAYD adoption. but large social benefit from PAYD, a booster shot from the government may be needed for anFirst, enact regulatory and legal reforms to insurance firm to offer it, which may then pushpromote PAYD. At a minimum, states, where other firms to follow suit. Such a pilot programnecessary, should enact model legislation and would teach insurance firms, regulators, and theregulatory guidance permitting PAYD. For public about the feasibility and benefits of PAYD.example, legislation passed in Texas in 2001 gaveinsurers permission to offer cents-per-mile pricing Currently, funding for PAYD pilot programs hasfor vehicle insurance. The bill also required come from the federal Value Pricing Pilot (VPP)insurance companies to separately track and program, authorized under the Safe, Accountable,report the claim losses and premium revenues for Flexible, Efficient Transportation Equity Act: Amileage-based and time-based premiums (Texas Legacy for Users (SAFETEA-LU), from federalHouse Bill 45). Even in many states where PAYD funds distributed to states by formula, and from oth-is technically permissible it may be necessary for er government sources. The VPP program providesthe legislature to enact legislation clarifying that $12 million a year for pilots related to congestionPAYD is permitted in order to signal that insurance pricing, innovative parking programs, PAYD, andregulators will look favorably on PAYD rate other pricing strategies to reduce congestion. Ofsubmissions and reconcile potential conflicts with these funds, a minimum of $3 million a year must beexisting regulations that were written for lump- used for pricing projects not involving tolls, such assum rate plans. As already discussed, in California, PAYD. PAYD grants have ranged between $1 mil-for example, a bill and rulemaking process is aimed lion to $2 million. In March 2007, King County,at permitting insurance firms to vary per-mile Washington was awarded $1.9 million in Federalpremiums depending on whether the odometer Highway Administration (FHWA) discretionaryreadings are verified or self-reported. funds for its PAYD pilot, which also received $1.2 million in state and local funds. Unigard, the win-While state regulatory action is needed, states ning bidder, is matching these grants, investing $3.3may undervalue the importance of PAYD because million in the pilot. In 2006, Progressive was award-some of the benefits, like energy security and cli- ed $1.3 million in federal, state, and local fundingmate change, are national (indeed, global). If state for a mileage-based (but not true PAYD) pilot inregulation continues to be a barrier to firms offer- North Central Texas. This study monitored the re-ing PAYD and states do not address the issue, the sponse of three thousand customers.federal government should encourage states to act w w w.hamiltonproject.org   |    J U LY 2 0 0 8 21
  24. 24. Pay- As- Y ou-Drive A uto InsuranceCongress should reauthorize the VPP when the Given that the monitoring costs may exceed the ex-SAFETEA-LU bill is up in 2009, and should add $3 pected benefit of PAYD to insurance firms but aremillion a year targeted solely to funding true PAYD much smaller than the social benefit, governmentpilot programs. A true PAYD insurance program is should respond to the classic market failure arisingone where the unit of exposure is equal to one ve- from PAYD’s positive externalities by providing ahicle mile, and the total units are multiplied by the temporary tax credit to any insurance company forpolicy rate to produce the policy premium.12 The each customer it signs up to a true per-mile pre-new, targeted funding would be enough to support mium. As discussed in §3 (B.i), monitoring costs cana new pilot in five to ten large metro areas over total well over $100, while the benefit to the firm isthe life of the program, which we expect to last five likely to be quite small. Thus, we would propose ayears. $100 per-vehicle tax credit for each new true per- mile policy, which would be phased out once 5 mil-Third, address the market failure around mon- lion vehicles are covered by PAYD policies. Givenitoring costs. The government should require that there are roughly 225 million light-duty ve-that odometer readings be performed as part of re- hicles (cars, vans, pickup trucks and SUVs) in thequired safety and emissions inspections or by cer- United States, the tax credit would be eliminatedtifying vehicle service businesses in other states to once 2 percent of all premiums were of the PAYDperform odometer readings. Odometer audits are variety. The $100 tax credit would be available toinexpensive and can be performed while the vehicle any firm, although customers involved in a feder-is undergoing other servicing (Litman 2007a). If ally funded pilot program would not qualify. Thethese audits were performed along with traditional tax credit could be tailored, in theory, to give privateservicing, the time inconvenience to customers firms incentives to set up their programs in a man-would be minimal. The trouble is that today there ner likely to give the best information to consum-is no infrastructure in place to certify and regulate ers and elicit a reduction in VMT. For example, theproviders of odometer readings to ensure their credit could increase depending on how many timescredibility. a year the customer is billed, or if the firm’s moni- toring technology informs the driver in real timeStill, even with these reforms, trying to read each about how his driving is affecting his premium.vehicle’s odometers in a verifiable manner mayprove difficult. Technological options are more We cap the credit at roughly 2 percent of all vehiclescostly but are less burdensome on individuals be- because we believe that while this new product needscause they obviate the need to bring vehicles for a push, once PAYD is offered its dynamic effects onodometer readings. They are also likely to elicit a the market will quickly induce more and more driv-stronger behavioral response because the price sig- ers to adopt this form of insurance pricing. Presum-nals would be sent more frequently to the driver, ably, the first 2 percent of customers signing up toas discussed in §3 (A). Some technology options PAYD policies will be low-risk, low-mileage driverswould allow risk-adjusted mileage pricing, based on that have a financial incentive to do so. But oncefeatures such as time of day, driving area, and driver these drivers are out of the per-year risk pools, aver-actions related to speed, acceleration, and braking age annual accident costs for the remaining drivers(though these may raise more privacy concerns). In will increase, thus so too will annual premiums, overaddition to the technology costs, firms face high time. This will give even more drivers a financialdevelopment costs associated with switching from incentive to switch to PAYD, which will further in-lump-sum to per-mile premiums, whatever forms crease costs for those remaining under traditionalof mileage monitoring they choose.12. Insurers would have the option of varying the policy rate by speed, time of day, or other information gathered about the exposure unit.22 THE H AMILTON PROJECT    |   the broo king s ins titution
  25. 25. Pay-A s -Yo u - D r i v e A u t o In s u r a n c epolicies, leading a few more drivers to adopt PAYD.A virtuous circle will soon develop that leads nearlyall drivers to opt for PAYD policies. Another ben-efit of this dynamic is that PAYD adoption shouldquickly grow without having to make PAYD man-datory. Customers will always have the option tokeep their current premiums, but over time moreand more drivers will have a financial incentive toswitch to PAYD.The roughly $515 million cost of both the pi-lot program expansion and tax credits is less than1 percent of the annual social benefits expectedfrom PAYD, to which we turn in the next section.Moreover, even this small amount of governmentspending would not require new revenue. As dis-cussed in §5 (B iii), the federal government wouldsave roughly $1.4 billion annually because feweraccidents would mean less Medicaid and Medicarepayments to accident victims and less lost tax rev-enue from reduced productivity of incapacitated orfatally injured workers. w w w.hamiltonproject.org   |    J U LY 2 0 0 8 23
  26. 26. Pay- As- Y ou-Drive A uto Insurance5. Impacts, Costs, and BenefitsT he net social benefits if PAYD were universally We extend the previous analyses by using data on adopted would be approximately $50 to $60 vehicles, driving, and household characteristics billion a year, mostly from reduced accidents from the 2001 National Household Transporta-and congestion, as well as from reduced local pollu- tion Survey (NHTS), which allows us to estimatetion, carbon emissions, and oil dependence. Insur- the driving reduction and cost impact of PAYD atance premiums would decline for almost two-thirds the vehicle and household levels. The national dataof drivers, since a minority of high-mileage drivers set includes observations from 21,374 householdsare responsible for the majority of miles driven. In with full information on annual mileage and fuelthis section, we discuss in more detail the expected economy for each vehicle owned by that house-social and individual benefits, and consider who hold. We assume PAYD is available for light-dutymight be harmed. vehicles (cars, vans, pickup trucks, and SUVs). We ignore buses and heavy trucks because their insur-A. Estimating the National Impact of ance companies probably have more informationPAYD about their mileage, and thus their premiums prob- ably already closely reflect mileage. Light-duty ve-Two previous studies have quantified the effect that hicles were responsible for 92 percent of all vehiclea switch to per-mile premiums would have on driv- miles in 2006 (U.S. Department of Transportationing and welfare. Edlin (2003) uses insurance premi- [DOT] 2006, Table VM-1). In total, 41,672 vehiclesum data to calculate the average per-mile insurance are included in our sample.premium in each state. He inserts per-mile premiumestimates into a driving and accident model to pre- We combine the information on annual mileage anddict driving reduction and accident savings state by fuel economy from the NHTS data set with otherstate. Using data from the late 1990s, he estimates driving-related data to calculate the expected driv-driving would decrease by about 10 percent nation- ing reduction from PAYD. We follow Parry (2005)ally, resulting in up to $20.5 billion in benefits from in modeling the driving response from PAYD.reduced accidents and congestion (updated to 2007 Drawing on gasoline and mileage demand elastici-dollars). Parry (2005) builds a general equilibrium ties from previous literature, Parry assumes per-milemodel that simplifies the analysis by aggregating insurance premiums as high as a vehicle’s per-mileall consumers into one representative agent. He in- fuel costs would reduce that vehicle’s mileage by 15cludes a more comprehensive set of driving exter- percent (for details, see appendix). As explained ear-nalities: carbon emissions, oil dependence, and local lier, only the collision, liability, and uninsured mo-pollution, in addition to accidents and congestion. torist portions of auto insurance premiums wouldParry uses gas price and fuel economy data from be expected to switch to per-mile rates. For state-the first half of this decade and borrows Edlin’s es- level data on these rates, we use a more recent ver-timate of national average per-mile premiums. He sion of the data that Edlin (2003) analyzed from thefinds that a switch to PAYD would decrease driving National Association of Insurance Commissionersand fuel consumption by 9.1 percent, resulting in (2007, Tables 1 and 2).13 In 2001, these portions ofa $20.5 billion welfare gain (updated to 2007 dol- the premium accounted for 84 percent of averagelars).13. In the NAIC tables, premiums for uninsured motorist coverage are included in liability premiums.24 THE HAMILTON PROJECT    |   the broo king s ins titution

×